Performance marketers are being asked to hand algorithms the keys to campaign execution while proving that every dollar delivers measurable business value.
That tension is real. But the solution is not to fight automation or give up full control. The way forward is to gain clarity on the business outcomes you want to optimize with AI and identify where human control remains non-negotiable.
This balancing act took center stage during September’s MarTech Conference panel with Maria Corcoran, performance media manager at Jiffy.com; Anthony Tedesco, global head of performance media at Cisco Systems; and Jiaxi Zhu, Head of Analytics at Google, moderated by Christina Inge, CEO of Thoughtlight.
AI makes clear goals more important, not less
Giving up granular control over campaigns is inconvenient when you’re responsible for the bottom line.
Corcoran noted that adapting to AI requires going beyond basic campaign settings. Marketers need to ensure that the AI truly understands how the target audience, product catalog, website and surrounding content connect.
Zhu stressed that the key is to clearly establish priorities. Algorithms cannot maximize every metric simultaneously without trade-offs. The fundamental step is to define the primary objective.
“As long as this goal is achieved,” Zhu noted, whether it was achieved “with AI or not with AI” becomes a secondary question.
Tedesco highlighted a common obstacle: the final conversion event is not always the best signal to train an algorithm. Enterprise B2B companies care about pipeline and revenue, but these events often occur too infrequently to fuel data-hungry AI models. The sweet spot is identifying proxy signals that occur often enough to train the system while keeping campaigns aligned with high-value business outcomes.
Know when to let the algorithm take control
There are areas where delegating to AI makes strategic sense.
For Tedesco, real-time bidding is a great example. An algorithm evaluates thousands of contextual signals during a search auction much faster than any human operator can. Assembling creative assets is another: models can quickly test and identify which combination of text and images resonates best with a specific segment of users.
However, Corcoran’s experience with Google Performance Max shows why automation still requires careful monitoring.
Jiffy.com operates four distinct lines of business. When testing a Performance Max campaign, the tool generated a high overall ROI, but allocated budget disproportionately to a single line of business, which had not funded the initiative.
That test provided a valuable strategic lesson: Jiffy’s site architecture didn’t clearly differentiate its service lines for AI models. What started as a test campaign became an opportunity to improve how the brand’s digital infrastructure communicates with automated systems.
Don’t eliminate parameters that matter yet
As research evolves, measuring brand presence within AI-generated summaries is becoming essential. Tedesco shared that Cisco monitors AI visibility metrics to monitor how large language models interpret and cite their content.
At the same time, key performance metrics remain critical.
“There’s no need to reinvent the wheel,” Tedesco noted. Traditional funnel metrics continue to serve as reliable anchors even as new signals are introduced.
Corcoran relies on key B2C metrics like LTV:CAC and cost per acquisition, using artificial intelligence to streamline cross-channel data analysis, uncover attribution discrepancies, and evaluate how influencers or user-generated content impacts performance.
AI doesn’t require abandoning proven metrics – it offers clearer visibility into what drives them.
Beyond campaign management, AI’s most immediate value lies in eliminating repetitive administrative work.
Zhu highlighted that AI significantly reduces friction in fundamental analysis, problem solving, and campaign setup, allowing teams to focus on strategy and cross-functional leadership.
Tedesco emphasized ad traffic, a rules-based activity that can be transformed into a streamlined and intuitive process. He also sees huge potential in self-service analytics.
While custom data joins previously required SQL skills or weeks of waiting in analytics queues, natural language AI tools now surface that information in minutes.
Corcoran uses tools like Claude to unify financial data, advertising metrics, site analytics and sales data into cohesive reports. Its main goal was practical: eliminate three hours of daily reporting tasks.
By leveraging AI to handle complex tasks like performing n-grams or correlation analysis, performance marketers can expand their analytical capabilities without needing a degree in data science.
Don’t confuse AI adoption with AI success
Industry adoption is still in its early stages. A survey conducted at a conference revealed that 58% of attendees are experimenting with AI for performance analysis, 23% are exploring potential use cases, and only 11% have fully implemented the technology into their workflows.
Zhu cautioned against measuring success solely by tool adoption or platform logins. The true indicator of success is whether AI applications improve actual business outcomes. Establishing clear benchmarks before testing ensures you only scale what works.
Provide AI with guardrails, not unlimited control
Managing automation comes down to setting limits.
“You need to find the balance between automation and autonomy that makes sense for your company,” Tedesco advised.
Think of campaign architecture as creating guardrails. Establishing a clean data taxonomy provides AI systems with the structure they need to generate accurate information and execute workflows reliably.
Corcoran recommends taking a measured approach to live system integrations, choosing to use AI extensively for background analysis before giving automated tools direct access to active market-facing budgets.
Artificial intelligence is not replacing the discipline of performance marketing; It’s raising the bar. The core mission remains unchanged: reach the right audience, deliver meaningful messages and drive business growth. AI processes data faster, but marketers set direction, validate data, and define boundaries.
The most important strategic question is not just how AI changes campaign management, but how AI changes how customers interact with your company. This is where sustainable growth begins.
Watch the Martech September 2026 conference for free, on demand.
